1 /* { dg-do assemble { target aarch64_asm_sve_ok } } */ 2 /* { dg-options "-O2 -ftree-vectorize --save-temps" } */ 3 4 #define INDEX8 uint8_t 5 #define INDEX16 uint16_t 6 #define INDEX32 uint32_t 7 #define INDEX64 uint64_t 8 9 #include "strided_load_1.c" 10 11 /* 8 and 16 bits are signed because the multiplication promotes to int. 12 Using uxtw for all 9 would be OK. */ 13 /* { dg-final { scan-assembler-times {\tld1w\tz[0-9]+\.s, p[0-7]/z, \[x[0-9]+, z[0-9]+.s, sxtw 2\]\n} 6 } } */ 14 /* The 32-bit loop needs to honor the defined overflow in uint32_t, 15 so we vectorize the offset calculation. This means that the 16 64-bit version needs two copies. */ 17 /* { dg-final { scan-assembler-times {\tld1w\tz[0-9]+\.s, p[0-7]/z, \[x[0-9]+, z[0-9]+.s, uxtw 2\]\n} 3 } } */ 18 /* { dg-final { scan-assembler-times {\tld1d\tz[0-9]+\.d, p[0-7]/z, \[x[0-9]+, z[0-9]+.d, lsl 3\]\n} 15 } } */ 19